Nosocomial outbreaks due to amikacin-resistant tobramycin-sensitive Acinetobacter species: correlation with amikacin usage

Journal of Hospital Infection (1990) 15, 83-93

Nosocomial outbreaks due to amikacin-resistant tobramycin-sensitive Acinetobacter species: correlation with amikacin usage Y. Buisson,

G. Tran

Van Nhieu,* L. Ginot, P. Bouvet,? L. Driot and M. Meyran$

H. Schill,

Laboratoire de Biologie Clinique, H&ital d’Instruction des Arme’es du Val-deG&e, Paris, France, *Laboratoire de Microbiologic Mkdicale, Universite’ Pierre et Marie Curie, Paris, France, I-Service des Ent&obact&ies, Institut Pasteur, Paris, France and SLaboratoire de Bactt%iologie, H6pital d’Instruction des Arme’es Bkgin, Saint-Mande’, France. Accepted for publication

14 September 1989

Summary: Fifty-seven patients in the Val-de-G&e hospital were infected or colonized with amikacin-resistant, tobramycin-sensitive Acinetobacter spp. between January 1985 and December 1987. This resistance phenotype was attributed to the recently described 3’-0-aminoglycoside phosphotransferase (APH(3’)-VI), on the basis of substrate profile and DNA-DNA hybridization, and was mainly encountered in various biotypes of A. baumannii isolated from patients. It was also encountered in saprophytic A. johnsonii isolates from the hands of 11 healthy workers among the medical staff, which provided evidence for the dissemination of an epidemic gene among different biotypes and species of Acinetobacter. A retrospective epidemiological survey showed a significant correlation between amikacin consumption and case incidence in the wards where cross-infection had occurred.

Keywords: Acinetobacter; nosocomial aminoglycoside

outbreak;

amikacin

resistance;

3’-0-

phosphotransferase.

Introduction

Nosocomial infections involving Acinetobacter spp. have been of increasing importance in the last few years (Joly-Guillou & Bergogne-Berezin, 1985). These are aerobic non-fermentative Gram-negative coccobacilli, which can be isolated from environmental sources or human skin. Acinetobacter genus has been recently subdivided into 12 genospecies (Bouvet & Grimont, 1986), including A. baumannii which is often responsible for colonization and infection of patients in intensive care (Castle et al., 1978), in Correspondence Val-de-GrBce,

to: Y. B&son, 74, Bd de Port

01954701/90/010083

+ 11 so3.OQjo

L a b oratoire Royal, 75230

de Biologic Clinique, HBpital Paris cedex 05, France.

d’hstruction

0 1990 The Hospital

83

des Arm&s

Infection

du

Society

84

Y. Buisson

et al.

haemodialysis (Abrutyn et al., 1978), neurosurgery, and burn units (Bergogne-Berezin, Joly-Guillou & Vieu, 1987). These species are naturally resistant to cephalosporins and to kanamycin, and can frequently acquire plasmid-mediated resistance to penicillins, chloramphenicol and aminoglycosides (Murray & Moellering, 1979; Devaud et al., 1982). Several aminoglycoside modifying enzymes have been reported in Acinetobacter resistance to gentamicin, sisomicin and tobramycin SPP.7 conferring (Bergogne-Berezin et al., 1980). A mi k acin remains the most effective aminoglycoside against these organisms, as resistance to this antibiotic occurs with a relatively low frequency compared with other aminoglycosides. Enzymic resistance to this antibiotic depends either on the synthesis of 6’-N-acetyltransferase type 4, or of a new type of 3’-0-phosphotransferase (Lambert, Gerbaud & Courvalin, 1988). This latter mechanism, which has been characterized in various species of Acinetobacter, is typically associated with resistance to kanamycin but susceptibility to tobramycin. Since 1981, amikacin use has been increasing at the Val-de-Grlce hospital, and was employed as the aminoglycoside of first choice from 1982 to 1987. During this period, repeated nosocomial outbreaks of amikacin-resistant Acinetobacter baumannii occurred in different wards of the hospital. In this study, we investigated the relationship between amikacin use and the occurrence of amikacin-resistant Acinetobucter. The possible role of an in viva transfer of resistance, from naturally resident bacteria of human skin to infecting or colonizing strains, is also discussed. Materials

and methods

Epidemiological studies The military Val-de-Grace hospital is a nine-year old teaching institution. It has 480 beds distributed in 13 wards, including intensive care (ICU), neurosurgery (NSU), haematology, haemodialysis, radiotherapy and urology units, where nosocomial infections occur with relatively high frequency. Since 1985, hospital-acquired infections have been recorded by the microbiological laboratory. In this epidemiological survey, the case definition consisted of every in-patient infected or colonized by an amikacin-resistant, tobramycin-sensitive strain of Acinetobacter. A fingerprint method on agar containing amikacin (10 mg 1-l) and vancomycin (5 mg 1-i) was used to detect amikacin-resistant Acinetobacter hand carriage among medical and nursing staff. Identi)ication of isolates Acinetobacter isolates were identified using routine methods and classified into A. baumannii, A. haemolyticus, and A.johnsonii species according to the new taxonomic scheme (Bouvet & Grimont, 1986), using a 16-test system

Amikacin-resistant

Acinetobacter

outbreak

85

associated with tests for growth at 37”C, 41°C and 44°C. A complementary biotyping system allowed the identification of 17 biotypes (Bouvet & Grimont, 1987). Phage typing was performed at the phage typing centre of the Pasteur Institute, Paris. Antibiotic susceptibilities Susceptibility to antibiotics was tested by using the disk-diffusion method on Mueller-Hinton agar. Resistance or sensitivity phenotype was assigned from the inhibition zone diameter in the manner dictated by statement of the Antibiogram Committee of the French Society of Microbiology (Acar et al., 1986). Minimum inhibitory concentrations (MICs) were determined by using a twofold agar dilution method with a Steers-type replicator. Mechanism of resistance to amikacin Total DNA, prepared as described previously (Tran Van Nhieu & Collatz, 1988), was suitable for dot blot analysis. A 370 bp EcoRI-BamHI DNA fragment of the gene coding for the APH (3’)-VI was used as a probe. It was obtained from the pUC%derived plasmid pAT240 (Lambert, Gerbaud & Courvalin, personal communication), which was generously provided to us. Controls used for dot blot analysis were the APH(3’)-VI encoding A. baumannii strain BM2580, and its plasmid-cured derivative BM2582 (Lambert, Gerbaud & Courvalin, 1988). used to test for aminoglycoside Crude enzyme extracts were by the phosphocellulose paper binding phosphotransferase activity, technique using a-32P ATP (Haas & Dowding, 1975). Statistical analysis Results of the case-control survey and correlation of the case occurrence with amikacin use were analyzed by the Chi-square method according to Mantel-Haenzel and the coefficient of rank correlation of Spearman, respectively. Results

Epidemiological investigation Since 198.5, Acinetobacter was responsible for up to 10% of hospital-acquired infections. From March 1985 to August 1987, a total of 57 in-patients were infected or colonized with an amikacin-resistant, tobramycin-sensitive Acinetobacter sp. The epidemic curve showed three distinct outbreaks, with peaks occurring in August 1985, February 1986 and April 1987 (Figure 1). Patients were 52 males and 5 females, from 14 to 82 years old (mean 46.6, median 45). They were hospitalized in 8 different units when the index strain was first recovered, namely ICU (68%), NSU (14%), surgery and ophthalmology units (5%), urology, haematology,

Y. Buisson

et al.

. 1

3h

vlAMJJA$

1985 Figure 1. Monthly square represents amikacin-resistant phenotype.

I

1

i 13NC 1986

case-finding and quarterly amikacin use over the three-year period. One the first strain of Acinetobacter recovered from each patient: ‘6% tobramycin-sensitive Acinetobacter; 0 Acinetobacter with other resistance

radiotherapy and neurology units (2%). Underlying diseases mainly consisted of major trauma (n = 17), pneumonia (n = 1 l), and intracranial tumours often requiring assisted ventilation (n= 10). Time between admission to the hospital and first isolation of an amikacin-resistant Acinetobacter sp. varied from one to 81 days (mean 19 days). Among the 57 patients, 38 were colonized, often from several superficial sites such as skin, buccal and nasal mucosa, conjunctiva, and sites of insertion of devices such as drains, catheters, and tracheal cannulae (Table I). Thirteen other patients were found infected at the time of the first isolation, whereas six were colonized from 12 days to 7 months before infection. Infections consisted of two cases of septicaemia, six of meningitis (four after intracranial surgery), three respiratory-tract infections, four urinary tract infections, and four skin and wound infections. In two cases, fatal outcome was linked to the infection. The overall duration of carriage of the amikacin-resistant organism was less than one week for 24 patients, between

Amikacin-resistant Table

I. Origin

Origin

of specimens

Acinetobacter

87

outbreak

of specimens and clinical

significance of 127 isolates of amikacin-resistant tobramycin-sensitive Acinetobacter spp. Clinical

significance

of Acinetobacter

Colonization

Infection Blood Cerebrospinal fluid Pleural fluid Urinary tract Vagina Faeces Conjunctiva Respiratory tract Skin and wounds Drains and catheters

isolates

30

1 1

8 :

Total

one week and one month patients.

9

30 10 16

31

96

for 20 patients,

and more than one month

for 13

Antibiotic susceptibilities Almost all the amikacin-resistant Acinetobacter isolates in the 57 cases belonged to A. baumannii species (n = 56), with the exception of one isolate of A. haemolyticus. MICs of some antibiotics are shown in Table II. Most of the isolates were resistant to beta-lactams, except for ceftazidime and imipenem. In the case of aminoglycosides, resistance to amikacin was associated with resistance to kanamycin, gentamicin and sisomicin, but susceptibility to tobramycin.

Table

II.

Antibiotic

susceptibilities of 57 outbreak-associated Acinetobacter and one A. haemolyticus (one isolate per case)

isolates: 56 A.

baumannii Antibiotics

Ticarcillin Piperacillin Aztreonam Ceftazidime Kanamycin Gentamicin Tobramycin Amikacin Netilmicin Pefloxacin

MIC

(mg 1-i)

Range

Mode

8-4096 16256 4256 0.5-128 128-256 0.5-256 l-4 162.56 4-128 0.25128

256 256 64 25: 256 1 128 8 64

88

Y. Buisson

et

al.

Typing of A. baumannii isolates The 56 A. baumannii isolates were classified into three biotypes (1, 2 and 6). Phage typing allowed the separation of 44 isolates into 6 phage types; 12 were either atypical or insensitive (Table III). A correlation was observed between biotypes 2 and 6, and phage types 75 and 17, respectively. These biotypes were prevalent among the 56 A. baumannii isolates. The biotype 1 included phage types 86, 88 and 89. The incidence of the three biotypes in ICU, NSU, and other units is represented on Figure 2. Biotype 6 was prevalent during the first and the last outbreak in the ICU, whereas biotype 1 was mainly isolated between December 1985 and April 1986, during the second outbreak, in the different wards. Biotype 2 was first isolated in the NSU in May 1985, and was subsequently found in the ICU and other units (Figure 2). Environmental reservoirs and human carriers Analysis of inanimate sites did not permit the establishment of any However, the fingerprint analysis environmental source of infection. performed from March to July 1986 led to the isolation of an amikacin-resistant Acinetobacter from the hands of 11 staff workers out of 131 sampled. All these isolates belonged to A. johnsonii species and harboured the amikacin-resistant, tobramycin-sensitive phenotype. Mechanism of the amikacin-resistance phenotype Eleven isolates of A. baumannii, representing the three different biotypes recovered over the 30-month period, and one strain of A. haemolyticus were subjected to dot blot analysis, using an internal fragment of the previously described APH(3’)-VI gene. Total DNA prepared from these isolates showed positive hybridization (Figure 3). Three isolates of A. johnsonii,

Table

III.

Correlation

between phage-types

and biotypes of 59 outbreak-associated isolates

Biotypes

Phage types 1

of A. baumannii

2

6 1

75 86 88 89 Atypical Insensitive

10 4 5 1

8

23

5 2

A. baumannii

Amikacin-resistant

Figure 2. Occurrence of species and biotypes different wards over the 30-month period. biotype 6; q A. haemolyticus. Line 1: ICU,

Acinetobacter

outbreak

89

of outbreak-associated Acinetobacter spp. in the q A. baumannii biotype 1; m biotype 2; q line 2: NSU, line 3: other wards.

which were isolated from the fingers of the nurses and the physicians of the NSU, the ICU and the ophthalmology unit during the same periods, also showed hybridization in the same experiment (Figure 3). .Substrate profiles obtained from the crude extract of an A. baumannii and an A. johnson&’ were identical and consistent with an APH(3’)-VI (data not shown). These data were consistent with the dissemination of the same gene coding for an APH(3’)-VI conferring resistance to amikacin among A. baumannii and A. johnsonii strains. Correlation between resistance and amikacin usage Since its introduction into the hospital pharmacy, in April 1980, amikacin had become the aminoglycoside used most extensively. At the outset of the epidemic in 1985, amikacin was prescribed in every case of first,-line antibiotic therapy, mainly in combination with cefotaxime, and especially in the ICU.

Y. Buisson

90 I

2

3

4

et al. 5

6

7

c

,Figure 3. Dot blot analysis using an APH(S’)-VI encoding gene probe. Lanes A, B, and C: DNA corresponding to Acinetobactw isolated in 1985, 1986, and 1987, respectively. Lane A, 2, lane B, 2, 3: A. baumannii biotype 1. Lane A, 3-5, lane B, 1, and lane C, 1: A. baumannii biotype 2. Lane A, 6, lane B, 4 and lane C, 2: A. baumanniz’ biotype 6. Lane C, 3 and 4-6: A. haemolyticus and A. johnsonii, respectively. Lane A, 1 and lane C, 7: BM2582 (negative control) and BM2580 (positive control), respectively.

A retrospective case-control study was carried out, including 56 cases as defined earlier. Controls consisted of 104 non-matched in-patients, selected in the corresponding wards on being infected or colonized with an A. baumannii strain harbouring any other resistance phenotype, during the same period. A highly significant correlation (p < 0.001) was found between previous treatment with amikacin and the occurrence of amikacin-resistant, tobramycin-sensitive A. baumannii isolation (Table IV). Following the first outbreak, a restriction of amikacin use was instituted which led to a significant decrease in the case incidence during the year 1986. At the outset of the next outbreak, in February 1987, amikacin consumption exceeded 400 g per quarter (Figure 1). The calculation of the coefficient of ranks of Spearman (Table V), with a delay between cause and effects of either 2 months (r’ = 0.745, p < 0.05) or 3 months (r’ = 0.85, p < O-01), permitted the establishment of a significant correlation between quarterly amikacin consumption and case incidence. There was no significant correlation between quarterly amikacin consumption and the incidence of infections or colonization due to all Acinetobacter isolates (Table V).

Amikacin-resistant Table

IV.

Effect

Acinetobacter

of amikacin therapy during the month preceding the first outbreak-associated A. baumannii (case-control survey)

Previous amikacin therapy

Isolates

isolates

Chi square = 15.9 (p < 0.001); *Ami-R= amikacin resistant;

isolation

Other phenotypes

17 39

7 97

56

104

Odds ratio = 6.0; Confidence tob-S = tobramycin sensitive.

of an

of A. baumannii

Phenotype ami-R, tob-S*

Yes No Total

91

outbreak

interval

= 1.9; 18.6.

Table V. Correlation between quarterly consumption of amikacin and monthly incidence of Acinetobacter isolationfromJanuary 1985 toJune 1987 (numbers mean values of coeficient r’ of ranks of Spearman) Acinetobacters

Delay

between

amikacin

0 ami-R, tob-S phenotype All phenotypes *Significant

value

(pi

0.05);

0.13 -0.515 tsignificant

value

consumption (months)

and case incidence

1

2

3

0.37 - 0.485

0.745* - 0.29

0.85t - 0.23

(p < 0.01).

Discussion

The emergence of a new resistance phenotype among nosocomial isolates is a clinically important event, as infections due to these organisms seem to occur with increasing frequency (Joly-Guillou & Bergogne-Berezin, 1985). was first amikacin-resistant, tobramycin-sensitive phenotype The characterized in isolates of Acinetobacter from several Parisian hospitals (Lambert, Gerbaud & Courvalin, 1988), and has been attributed to a new type of aminoglycoside 3’-0-phosphotransferase (APH(3’)-VI). We report in this epidemiological study the occurrence of strains of Acinetobacter resistant to amikacin via an APH(3’)-VI, isolated from infected or colonized patients over a 30-month period, in a hospital where amikacin was used as a first-line antibiotic. Typing of A. baumannii isolates that were responsible for infection or colonization of patients, revealed that biotype 6 had first spread in the ICU, then in the NSU and the other wards. The subsequent extension of this

92

Y. Buisson

et al.

mechanism of resistance to biotypes 2 and 1, as shown by the dot blot analysis, provided evidence for an in viva transfer of the APH(3’)-VI encoding gene. The finding of this type of resistance in A. johnsonii, which are normal saprophytes of human skin and their isolation from hands of medical staff, suggest that this species may have played an intermediary role in the transmission and dissemination by hand carriage of the amikacin-resistance gene to more pathogenic organisms. Further investigations would be necessary to characterize the potential role of the resident flora as a reservoir of transmissible resistance factors. The mechanism of transfer has not been elucidated. Plasmid analysis revealed that most of the isolates studied harboured several plasmids, but in our experiment efforts to transfer the amikacin resistance by conjugation were unsuccessful. A correlation has been established between amikacin use and the occurrence of amikacin-resistant Acinetobacter. Also, further analysis showed a larger case finding in the ICU, where amikacin was most intensively consumed,, and the establishment of previous treatment with amikacin as an individual risk factor. These data are consistent with the role of amikacin as a selective agent for resistant Acinetobacter. An increase in the frequency of amikacin-resistant Gram-negative bacilli linked to the use of amikacin has already been observed (Levine et al., 1985). However, in other prospective surveys, the frequency of resistance to amikacin did not change significantly, despite intensive use of this antibiotic, and the frequency of resistance to gentamicin and to tobramycin tended to decrease (Price et al., 1981, Moody et al., 1984). From our study, the decrease of case incidence from April 1985 to August 1986, which correlated with a diminution of amikacin use, has to be discussed. The introduction of infection control policies in March 1985, mainly consisting of patient isolation and repeated hand washing by medical staff, could partly explain the phenomenon observed during this period. However, the efficiency of such policies is difficult to estimate, as it is likely to vary among the different wards. This could explain the unexpected decrease of case incidence from April to October 1987, concomitant with an increase in amikacin use. In that period, amikacin had been recently selected for therapeutic schedules of sepsis occurring in granulocytopenic patients and was mainly consumed in the haematology unit where strict isolation of patients prevented cross-infection more efficiently. The emergence and spread of amikacin-resistance determinants among nosocomial organisms such as Acinetobucter is an alarming problem. From this epidemiological study, it appears that continuous surveillance, with restriction of amikacin use when necessary, together with improved hygiene practices, may limit the occurrence of resistant organisms, especially in wards where cross-infection is frequent.

Amikacin-resistant

Acinetobacter

outbreak

93

We are grateful to J.F. Vieu for typing of Acinetobacter strains. This work was supported in part by a grant from the Caisse Nationale de L’assurance Maladie des Travailleurs SalariCs (grant no. 87-3-22-07-E to E. Collatz).

Abrutyn, E., Goodhart, G., Roos, K., Anderson, R. & Boxton, A. (1978). Acinetobacter calcoaceticus outbreak associated with peritoneal dialysis. American Journal of Epidemiology 107, 328-335. Acar, J. F., Goldstein, F., Menard, R. & Bleriot, J. P. (1986). Strategies in aminoglycoside use and impact upon resistance. The American Journal of Medicine 80 (Suppl. 6B), 82-87. Bergogne-Berezin, E., M. L., Moreau, N. & Legoffic, F. (1980). Joly, Aminoglycoside-modifying enzymes in clinical isolates of Acinetobacter calcoaceticus. Current Microbiology 4. 361-364. Bergogne-Berezin, E., Jldly-Guillou, M. L. & Vieu, J. F. (1987). Epidemiology of nosocomial infections due to Acinetobacter calcoaceticus. Yournal of Hosbital Infection 10. 105-l 13. Bouvet, P. & Grimont, P. (1986). Taxonomy of ths genus &zine&bacterwith the rkcognition of Acinetobacter baumannii sp. nov., Acinetobacter haemolyticus sp. nov., Acinetobacter johnsonii sp. nov., and Acinetobacter junii sp. nov., and amended descriptions of Acinetobacter calcoaceticus and Acinetobacter lwoji. International Journal of Systematic Bacteriology 36, 228-240. Bouvet, P. & Grimont, P. (1987). Identification and biotyping of clinical isolates of Acinetobacter. Annales de Microbiologic (Institut Pasteur) 138, 569-578. Devaud, M., Kayser, F. & Bachi, B. (1982). T ransposon-mediated multiple antibiotic resistance in Acinetobacter strains. Antimicrobial Agents and Chemotherabv 22. 323-329. Joly-Guillou, M. L. & Bergogne-Berezin, E. (1985). E;olution d’Acinetoba&r calcoaceticus en milieu hosnitalier de 1971 B 1984. La Presse Medicale 14. 177-l 81. Lambert, T., Geibaud, G. & Courvalin, P. (1988). T ransferable amikacin resistance in Acinetobacter spp. due to a new type of 3’-aminoglycoside phosphotransferase. Antimicrobial Agents and Chemotherapy 32, 15-l 9. Levine, J., Maslow, M., Leibowitz, R., Pollock, A., Hanna, B., Schaefler, S., Simberkoff, M. & Rahal, J. (1985). Amikacin-resistant gram-negative bacilli: correlation of occurrence with amikacin use. The rournal of Infectious Diseases 139, 371-375. Moody, M., C. A. de Jongh, S. C. Schimpff, & G. L. Tillman. (1982). Long-term amikacin use: effects on aminoglycoside susceptibility patterns of gram-negative bacilli.Journal of the American Medical Associatian 248, 1199-l 202. Murray, B. & Moellering, R. (1979). Aminoglycoside-modifying enzymes among clinical isolates of Acinetobacter calcoaceticus subsp. anitratus (Herellea vaginicola): explanation for high-level aminoglycoside resistance. Antimicrobial Agents and Chemotherapy 15, 190-199. Price, K. E., Kresel, P. A., Farchione, L. A., Siskin, S. B. & Karpow, S. A. (1981). Epidemiological studies of aminoglycoside resistance in the USA. Journal of Antimicrobial Chemotherapy 8 (Suppl.A.), 89-105. Tran Van Nhieu, G. & Collatz, E. (1988). Heterogeneity of 6’-N-acetyltransferases of type 4, AAC(6’)-4, conferring resistance to amikacin and related aminoglycosides in enterobacteriaceae. Antimicrobial Agents and Chemotherapy 32, 1289-1291.